BEFORE THE
POSTAL RATE COMMISSION
WASHINGTON, D.C. 20266-0001
POSTAL RATE AND FEE CHANGES, 2000
Docket
No. R2000-1
DIRECT TESTIMONY OF SANDER A. GLICK
ON BEHALF OF THE ASSOCIATION FOR POSTAL COMMERCE
AND
MAIL ADVERTISING SERVICE ASSOCIATION
Communications
regarding this document
should be served on
Ian D. Volner
N. Frank Wiggins
Venable, Baetjer, Howard & Civiletti, LLP
1201 New York Avenue, N.W.
Suite 1000
Washington, DC 200053917
Dated:
May 22,200O
TABLE OF CONTENTS
AUTOBIOGRAPHICAL
I. PURPOSE
SKETCH . .. . .. . .. . . .. . .. . .. . . .. . .. . . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . . .. . .. . . .. .. . . .. ii
AND SCOPE OF TESTIMONY
II. DROPSHIP
DISCOUNTS
COST SAVINGS
. . .. . .. . .. . . .. . .. . .. . .. . .. . .. . .. . .. . .. . . .. . .. . .. . .. . .. 1
SHOULD REFLECT
ALL DESTINATION
ENTRY
. . . .. . .. . . .. .. . .. . .. . .. . .. .. . .. . .. . .. . . .. . .. . . .. . .. . .. . .. . . .. . .. . .. . .. . .. . .. .. . . .. . .. . . .. . .. . .. .. . 2
A. The Postal Service’s Prooosal Passes Throuoh Onlv About 75 Percent of
Destination
Entrv Cost Savinss . . .. . .. . .. . .. . .. . .. . . .. . .. . . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . . .. . .. . .. . .. 2
B. Full Passthrouohs
Send the Appropriate
C. Full Passthrouohs
Are Consistent With Recent Commission
D. PostCorn’s Proposed
100 Percent Passthrouohs
III. THE VALUE OF AUTOMATION
THE POSTAL SERVICE
Decisions . . .. . . 3
Are Appropriate
. .. . .. . .. . .. . . 5
IS MUCH HIGHER THAN ESTIMATED
BY
.. . .. . .. . .. .. . .. . .. . .. . .. . .. . .. . . .. . .. . . .. . .. . .. . .. . .. . .. .. . . .. . .. .. . . .. . .. . .. . .. . . .. . 6
A. Incorrect and Inconsistent
to Understated
Pricina Sionals .. . .. . .. . .. . . .. . .. . .. . .. . .. . . 3
Assumptions
in the USPS Flats Cost Model Lead
Cost Savinas . .. . .. . .. . .. .. . . .. . .. . .. . . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. .. . . . .. . .. . .. . .. . .. 8
B. The Postal Service lanored the Cost Savinos That Result Directlv From
Automation
C. Barcoded
Reouirements
Pertainino to Address Qualitv . .. . .. . .. . .. . .. . .. . . .. . .. . .. . .. . 16
Flats Will Facilitate the Postal Service’s Flats Automation
Prooram.. .........................................................................................................
ATTACHMENT
A. SUMMARY
OF POSTCOM
PROPOSED
17
RATES.. ............ .20
1
AUTOBIOGRAPHICAL
SKETCH
My name is Sander A. Glick. I co-manage
2
3
at Project,Performance
4
Virginia.
5
and public sector clients.
6
Program Manager.
7
economic and cost issues for mailer associations,
6
and the Department
9
Corporation
the Economic Systems practice
(PPC), a consulting
PPC provides economic and technology
firm based in McLean,
consulting
services to private
I joined PPC in 1994 as an Analyst and am now a
Since joining the firm, I have worked on a number of
the Department
of Defense,
of Energy.
In Docket No. R97-1, I testified regarding the fee for Qualified Business
IO
Reply Mail (QBRM) and the appropriate
11
to mail classes and subclasses.
12
Magazine
13
methods and joint Postal Service/Industry
14
for the Recording
15
appropriate
16
flats and parcels.
17
Advisory Committee’s
16
industry observer on the MTAC Package Integrity Study.
19
I attended the Maxwell School of Citizenship
Publishers
method for distributing
rural carrier costs
In this case, I am also testifying for the
of America (MPA) regarding the Postal Service’s costing
Industry Association
method for determining
efforts to reduce Test Year costs and
of America (RIAA) regarding the
the cost difference
I am an industry representative
between Standard (A)
on the Mailers’ Technical
(MTAC) Package Integrity Work Group and was an
and Public Affairs at
20
Syracuse
University, where I received a Masters of Public Administration
21
and Carleton College, where I received a Bachelors
Degree, magna cum laude,
22
in Physics in 1993. I am a member of the American
Economic Association
23
the System Dynamics Society.
ii
in 1994,
and
1
I. PURPOSE
AND SCOPE OF TESTIMONY
In this testimony,
important attributes.
I propose a rate design for Standard (A) with two
First, it provides mailers with the appropriate
prepare their mailings in a way that minimizes the combined
transportation
mail processing
and
cost to mailers and the Postal Service in the Test Year. Second, it
encourages
mailers to barcode flats, which will facilitate the Postal Service’s
automation
program.
My proposed
9
incentives to
cost savings.
rate design passes through 100 percent of destination
entry
This rate design will not only increase the amount of Standard (A)
IO
mail that is dropshipped,
II
presented
12
describe in my testimony for MPA, increased
13
of bundles that will break in the Test Year, which in turn will lower Postal Service
14
costs.
15
handle than similarly presorted sacks.
16
but will also increase the amount of mail that is
on pallets (see witness Schick’s (PostCorn, et al.-T-2) testimony).
Further, as described
palletization
As I
will reduce the number
by witness Unger, pallets are much less costly to
USPS-ST-43
at 4-5.
Second, while I believe that witness Moeller’s (USPS-T-35)
discounts for Standard (A) flats are reasonable,
proposed
17
automation
16
flawed cost studies.
19
his discounts are understated.
20
considerably
21
them to be. Also, the Postal Service’s flats cost model ignores the value that
22
barcoded flats will provide to the Postal Service’s flats automation
23
large portion of the savings that result directly from automation
24
pertaining to address quality. While these benefits are hard to quantify, they are
25
high and justify a passthrough
26
savings in this case. The remainder of my testimony provides detail to support
27
these points.
Specifically,
they are based upon
the modeled cost savings upon which he bases
Thus, the proposed passthroughs
are a
smaller proportion of costs saved than witness Moeller presents
program and a
requirements
of more than 100 percent of the quantifiable
cost
2
1
Section II of my testimony summarizes
the Postal Service’s proposed
2
dropship discounts,
estimates dropship discounts
based on full passthroughs,
3
and describes
precedents
4
cost savings.
Section Ill discusses
5
understated
6
which the Postal Service understated
7
describes the value that barcodes will provide to the Postal Service’s flats
0
automation
for passing through 100 percent of destination
entry
several reasons why the Postal Service has
the cost savings that result from barcoding,
quantifies the degree to
the value of automation
mail, and briefly
program.
9
10
II. DROPSHIP
11
COST SAVINGS
12
13
A. The Postal Service’s
14
of Destination
15
DISCOUNTS
Proposal
REFLECT
ALL DESTINATION
Passes Through
Only About
ENTRY
75 Percent
Entry Cost Savings
In this docket, witness Moeller proposes passthroughs
16
(DBMC) and destination
17
respectively,
ia
passthroughs
19
for destination
20
Route (ECR) subclass.
21
SHOULD
of destination
BMC
SCF (DSCF) cost savings of 73 percent and 77 percent,
for the Standard (A) Regular subclass (USPS-T-35
at 15) and
of 73 percent for DBMC, 77 percent for DSCF, and 77.5 percent
delivery unit (DDU) entry for the Standard (A) Enhanced
Carrier
Id. at 27.
In contrast, I propose 100 percent passthroughs
22
cost savings because 100 percent passthroughs
23
signal to mailers and, as described
24
efficient mailstream.’
of all destination
send the appropriate
entry
price
by witness Schick,’ will result in a more
Also, full passthroughs
of destination
entry cost savings
‘Witness Schick notes that increased dropship discounts will increase palletization as well as the
volume of mail that is dropshipped. Data provided by witness Crum (USPS-T-27) further support
this point. Specifically, Crum shows that approximately 72 percent of Standard Mail (A) pounds
entered at destination BMCs and 68 percent of Standard Mail (A) pounds entered at destination
SCFs is prepared by mailers on pallets, whereas only 3 percent entered at origin associate
offices, 20 percent entered at origin SCFs. and 38 percent entered at origin BMCs are prepared
on pallets. USPS-T-27, Attachment C, Table 6.
‘Increased palletization will reduce Postal Service costs including container handling costs.
USPS-ST-43 at 4-5.
3
1
are more consistent with recent Commission
2
(A) and other mail classes.
recommendations
for both Standard
3
4
B. Full Passthroughs
Send the Appropriate
Pricing
Signals
5
In Docket No. R97-1, witness Bernstein described
6
100 percent of cost savings, an example of efficient component
7
sends the theoretically
8
9
IO
11
12
13
14
15
16
17
10
19
20
21
22
23
24
why passing through
pricing (ECP).
correct price signal:
...any activity that can be performed by more than one agent
should be performed by the most efficient (least cost) agent. In the
case of postal services, the principle of Efficient Component
Pricing can be applied to the establishment of a discount granted
to mailers for performing some task that would otherwise be
performed by the Postal Service, such as mailer presorting
instead of Postal Service sorting. ECP minimizes the total cost of
providing mail service, where the total cost is the sum of the
Postal Service’s cost plus the mailer’s cost of worksharing (known
as a user cost) if the mailer chooses to workshare. Under ECP,
the price difference between a non-workshared mail category and
its workshared component should equal the difference between
the Postal Service costs of the non-workshared and workshared
mail category. Docket No. R97-1, USPS-T-31 at 72-73.
In those situations where the cost for mailers to perform an activity is
25
significantly
different than the cost for the Postal Service to perform it, small
26
deviations from ECP will not have a large impact on mailer behavior.
27
Dropshipping,
26
indicates, increasing dropship discounts will increase palletization
29
dropshipping.
however, is not one of those activities.
As witness Schick
and
30
31
32
C. Full Passthroughs
Commission
Are Consistent
precedent
With Recent Commission
also indicates that destination
33
higher than those proposed
34
R97-1, the Commission
35
percent for Standard (A) mail. Op. R97-1 at V-431.
36
Commission
37
savings for all but the non-transportation
recommended
entry passthroughs
by witness Moeller are appropriate.
recommended
Decisions
In Docket No.
destirration entry passthroughs
100 percent passthroughs
Furthermore,
of 85
the
of destination
entry cost
portion of the DDU entry cost avoidance
4
1
for Periodicals
2
in that case.
Regular mail3 and nearly 100 percent for Standard
Id. at V-538 for Periodicals
In Docket No. MC951,
3
and V-488-495
the USPS proposed
all destination
5
(ECR) subclass and 95 percent for Standard (A) Regular.
6
153. The Commission
7
subclasses.
Id. at V-250.
6
passthrough
of the transportation
9
estimated for DSCF and DDU entry Periodicals
10
consideration
for Parcel Post.
100 percent passthroughs
4
entry cost savings for the newly proposed
recommended
(B) parcel post
Enhanced
In addition, the PRC recommended
and non-transportation
Carrier Route
Op. MC951
100 percent passthroughs
of
at V-l 52-
for both
a 100 percent
cost avoidances
Regular mail (again, excluding
of the unzoned editorial rate). Id. at V-141-142.
Table 1 presents a summary of the Commission-recommended
11
12
passthroughs
of destination
entry cost savings for the major mail subclasses
13
Docket Nos. MC95-1 and R97-1.
14
cost savings for Periodicals
15
passthroughs
16
mail.
Note that except for the passthrough
in
of DDU
Regular mail in Docket No. R97-1, these
are all higher than the Postal Service’s proposal for Standard (A)
3The editorial pound rate for Periodicals is flat. Therefore, the effective destination
for Periodicals mail is less than 100 percent.
entry discount
5
1
Table 1. Recent Commission-Recommended
Entry Cost Savings
2
Passthroughs
of Destination
for Major Mail Subclasses
Class/Subclass
MC95-1
R97-1
Periodicals
Regular
DSCF (PC/lb)
DDU
100/l 00’
100/l 00
100/l oo2
1oo/703
Standard (A)
Regular
DBMC
DSCF
1 oo4
100
855
85
Standard (A)
ECR
DBMC
DSCF
DDU
1006
100
100
85?
85
85
Standard (B)
Parcel Post
DBMC
DSCF
DDU
loo8
92
99
‘00. MC95-1 at V-141-142.
‘6;. R97-1 at V-537-536.
30n the pound rate, passthrough is 100 percent for transportation
savings and 70 percent for non-transportation
savings.
‘Op. MC951 at V-250.
sop. R97-1 at V-431.
‘Op. MC95-1 at V-250.
‘Op. R97-1 at V-431.
‘Op. R97-1 at V-466-493. DSCF discount of 45 cents per piece and
cost avoidance of 48.7 cents per piece; DDU discount of 72 cents per
piece and cost avoidance of 72.4 cents per piece.
3
4
5
D. PostCorn’s
Proposed
100 Percent
Table 2 presents a comparison
Passthroughs
Are Appropriate
of the per-piece and per-pound
discounts
6
that result from using the Postal Service’s reduced passthroughs
7
entry cost savings and the full passthroughs
a
demonstrates,
9
the discounts by 0.7 to 0.8 cents on a per-piece basis and three to four cents on
increasing the passthroughs
that I propose.
of destination
As this table
to 100 percent in all cases increases
6
1
a per-pound
basis. Also, it maintains the Postal Service’s proposed
2
DSCF differential
3
presents rates based upon these proposed
and increases the DSCF-DDU
differential.
DBMC-
Attachment
A
passthroughs.
4
5
6
Table 2. Comparison
and Passthroughs
of USPS and PostCom
for the Standard
Destination
(A) Regular
Carrier
Route:
Piece-rated
73
0.017
0.0235
DBMC
77
0.022
0.0289
DSCF
77.5
0.026
0.0357
DDU
Pound-rated
73
0.063
0.114
DBMC
77
0.106
0.140
DSCF
77.5
0.134
0.173
DDU
‘USPS-T-27 at 7: USPS-LR-I-166. wpl~wmm.xls, worksheet ‘drop.’
‘USPS-LR-I-166. wpl-comm.xls. worksheet “drop.”
Entry Discounts
and ECR Subclasses
100
100
100
0.024
0.029
0.036
0.007
0.007
0.008
100
100
100
0.114
0.140
0.173
0.031
0.032
0.039
7
a
Ill. THE VALUE OF AUTOMATION
9
THE POSTAL SERVICE
IS MUCH HIGHER THAN ESTIMATED
BY
10
11
In this case, witness Moeller states that the cost studies performed by
12
witness Yacobucci
13
are too large, so a reduction is warranted.”
14
proposes automation
15
order to mitigate the impact and to limit the percentage
16
rate cells, he proposes to maintain these discounts “at nearly 75 percent of their
17
current value.”
Id.
(USPS-T-25)
“clearly indicate that the automation
USPS-T-35
discounts
at 13. He therefore
discounts that are smaller than the current discounts but, in
changes for individual
7
1
Witness Moeller’s logic is flawed.
Incorrect and inconsistent
2
in the Postal Service’s flats cost model, USPS-LR-I-90,
3
by a significant
understate
assumptions
cost savings
amount:
4
.
5
The flats cost model overstates
nonbarcoded
6
FSM 881 accept rates for
flats.
7
.
6
The flats cost model fails to take into account the lower wage rates
that are paid to BCR/OCR
9
clerks as compared to keyers and clerks
who manually sort flats.
10
11
.
12
The flats cost model understates
13
secondary
14
the Test Year.
the proportion
of incoming
sorts that will be performed on flat sorting machines
in
15
.
16
The flats cost model uses inconsistent
assumptions
17
1000 keying and FSM 1000 BCR productivities.
16
data anomaly.
regarding
FSM
This results in a
19
20
Furthermore,
as described
by witness Lubenow (PostCorn, et al.-T-3), the
21
Postal Service has not quantified all of the cost savings that result from the
22
higher address quality that result directly from automation
23
while Postal Service witnesses
24
have on automated
25
Service has failed to quantify this value.
26
Therefore,
clearly understand
flat sorting productivity,
requirements.
Finally,
the effect that barcoded flats
Unger, Tr. 21/8275, the Postal
given these points and the rapidly changing flats processing
27
environment,
it is appropriate
to maintain automation
discounts at the level
28
proposed by witness Moeller.
29
reasons why the Postal Service’s model underestimates
30
automation
The remainder of this section discusses the
cost savings, models
cost savings using a revised version of the flats cost model, and
6
1
discusses
additional reasons why witness Moeller’s proposed automation
2
discounts should not be reduced.
3
4
A. Incorrect
5
Lead to Understated
6
and Inconsistent
Assumptions
in the USPS Flats Cost Model
Cost Savings
In this section, I first describe the reasons why the Postal Service’s cost
7
model understates
automation-related
a
accurate automation-related
9
flats cost model.
cost savings.
cost savings using an improved version of the USPS
While my corrections
environment,
I then estimate more
are focused on the Test Year flats
10
processing
the value of automation
will increase further in the
11
coming years as the Postal Service begins delivery point sequencing
12
barcoded flats.4
(DPS)
13
14
15
1. FSM 881 Accept
Rates
Based upon input from USPS Operations,
the Postal Service’s flats cost
16
model incorrectly assumes that the FSM 881 OCR accept rate for nonbarcoded
17
flats is between 80 and 90 percent5
ia
These accept rates are significantly
19
read rates that were cited by the Postal Service in its “Strategic Improvement
20
Guide for Flats Processing,”
21
LR-I-193 at 21.
22
Furthermore,
USPS-LR-I-90,
worksheet
“Accept Rates.”
higher than the 70 to 80 percent FMOCR
which was just issued in September
the FSM 881 accept rates for nonbarcoded
1999. USPS-
flats used in the
23
USPS flats cost model are similar to the average accept rates for FSM 881s in
24
“BCWOCR
25
nonbarcoded
26
USPS-LR-I-107,
mode,” a mode which processes
barcoded flats on the BCR and
flats on the OCR. USPS-LR-I-90,
Yrscrub.xls;
worksheet
Unger. Tr. 21/8175-8177.
“Accept Rates”;
Because accept rates
4Delivery Point Sequencing flats will increase USPS efficiency because the unit cost for
performing a manual incoming secondary sort for flat-shaped mail is more than four cents white
the unit cost of an incoming secondary sort on an AFSM 100 is approximately one cent. USPSLR-I-90. worksheet “Mailflow Model Costs”, column 21,
‘Note that the notation “FSM 661 OCR” in USPS-LR-I-90 denotes an FSM 661 in BCRlOCR
mode processing nonbercoded flats. Yacobucci, Tr. 50439. This is different than what “FSM
661 OCR” denotes in MODS. In MODS, “FSM 661 OCR” indicates en FSM 661 in BCR/OCR
mode processing all flats. Unger. Tr. 21/6175-6177.
9
1
are much higher for barcoded flats than for nonbarcoded
2
amounts to average accept rates for FSM 881s in BCR/OCR
3
the accept rates for nonbarcoded
4
secondary
5
Institutional
6
average accept rate for nonbarcoded
7
calculations,
a
FSM 881 BCR/OCR
9
approximately
flats is inappropriate.
FSM 881 BCR and FSM 881 BCR/OCR
Response to PostComlUSPS-ST43-6,
which were confirmed
mode as proxies for
Using actual incoming
accept rates for FY 2000,
I was able to calculate an
flats on the FSM 881 BCR/OCR.
by witness O’Tormey,
- Incoming Secondary
75 percent.
flats, using what
My
result in an average
accept rate for nonbarcoded
flats of
Tr. 21/8353-8354.
Since there is no reason to believe that the accept rate for other sorts
10
11
(e.g., outgoing primary) will be higher than that for the incoming secondary
12
and to be consistent with the “Strategic
13
Processing,”
I use this accept rate for all FSM 881 BCWOCR
14
nonbarcoded
flats. By increasing the effect of the presence of a barcode on the
15 ’ FSM 881 BCR/OCR
16
Improvement
sort
Guide for Flats
sorts of
accept rate, using this more accurate FSM 881 BCR/&R
accept rate for nonbarcoded
flats increases automation-related
cost savings.
17
ia
Table 3. FSM 881 BCR/OCR
FOut’
Scheme
going Primary
.AiT
.-4 .
Incoming Primary
Incoming Secondary
Accept
Rates for Nonbarcoded
USPS Flats Cost Model’
80%
80%
85%
88%
‘USPS-LR-I-90.
worksheet “Accept Rates.”
2. Average
Labor Rate
Flats
PostCom Model
75%
75%
75%
75%
19
20
Witness Kingsley indicates that the Postal Service’s flats cost model uses
21
22
identical labor rates for all clerks despite the fact that automated
23
operations
are staffed primarily by PS-04 and casual clerks while manual sorting
24
operations
are staffed by PS-05 clerks, and FSM keyers are PS-05 or PS-06
25
clerks.
26
FY 2001 average labor rates for each type of clerk.)
27
more likely to be sorted in automated
26
all operations
Kingsley, Tr. 5/1803-1804;
understates
Kingsley, Tr. 5/1840-1841.
operations,
automation-related
(BCR/OCR)
(Table 4 shows
Since automation flats are
usingan
cost savings.
average labor rate for
IO
Table 4. FY 2001 National
1
Average
Labor Rates’
Level
Rate
Casual
$11.49
PS-04
$27.41
PEG-05(Manual Sorting and Keying)
$31.41
PS-06 (Keying)
$32.93
‘Kingsley,
Tr. 511941.
2
3
To estimate worksharing-related
cost savings, the Postal Service should
4
develop operation-specific
wage rates in the next case and use them in their
5
worksharing
Lacking such information
6
data from witness Kingsley’s interrogatory
7
operation-specific
6
revised flats cost model (filed as MPA-LR-2)
9
Stralberg (l-W-T-1).
10
cost models.
responses
in this case, I have used
to develop approximate
wage rates for flat sorting operations.
In implementing
this adjustment,
I use these rates in the
which I developed with witness
I generally use PS-05 wage rates for
11
manual sorting, PS-05 and PS-06 wage rates for FSM keying, and casual and
12
PS-04 wage rates for OCR/BCR
13
wage rates to ensure that the weighted-average
14
operations,
15
Service’s model.
operations.
Then, I proportionally
adjust these
wage rate for all piece-sorting
$28.24, is consistent with the average wage rate in the Postal
USPS-LR-I-90,
worksheet
“Data.”
16
17
16
3. Incoming
Secondary
Operations
The incoming secondary
machine/manual
factors (IS factors) used in the
19
Postal Service’s flats cost model for FSM 881s and AFSM 100s understate the
20
extent to which Standard (A) flats will receive incoming secondary
21
sorters in the Test Year. Witness Yacobucci
22
percentages
23
secondary
24
flats not actually processed
25
25 at 16. For example, the IS factor of fifty percent in the USPS flats cost model
26
for AFSM 100s and FSM 881s for Standard (A) Regular flats, USPS-LR-I-90,
sorts on flat
defines IS factors as “the
of flats by machine type that flow to a machine for incoming
piece handlings that the machine actually processes.
on the machine are processed
The remaining
manually.”
USPS-T-
11
1
worksheet
“Data,” indicates that the Postal Service performs a manual incoming
2
secondary
sort on fifty percent of flats that meet all machinability
3
receive all other sorts on a flat sorter, and are sorted at plants that have
4
machines.
On cross examination,
5
requirements,
witness Kingsley justified the use of a fifty percent
6
IS factor for AFSM 100s and FSM 881s by first indicating that flats in small zones
7
would not be sorted on machines: “So, for flats, if we already know that they are
6
small zones, we aren’t going to put them on a flat sorter, that automatically
9
percent.”
is 30
Kingsley, Tr. 5/l 977.
Then she indicated that another twenty percent of flats are not sorted on
10
11
an FSM because they are rejected or cannot be handled by a machine.
12
Specifically,
13
cannot be handled by a machine, for one reason, and the pieces that are
14
rejected by a machine, for some reason, those are going to make up 20 percent
15
of the universe of flats in a test year. You add that to 30 and you get the 50, is
16
that right?” she responded
when asked, “So you are projecting that between the things that
“Approximately,
yes.” Kingsley, Tr. 5/1978.
While witness Kingsley’s argument that less than 100 percent of
17
16
machineable
flats will receive an incoming secondary
19
reasonable,
20
seventy percent (the percentage
21
routes), not fifty percent, for several reasons.
the appropriate
percentage
for Standard
sort on a flat sorter is
(A) Regular flats should be
of flats that destinate in zones with ten or more
First, witness Kingsley indicated that, at the end of the Phase I AFSM 100
22
23
deployment,
there will be no shortage of flat sorting machines.
Kingsley, Tr.
24
5/1631.
25
reason for manual flat sorting, will be alleviated by the Test Year. USPS-T-10
26
15; Kingsley, Tr. 5/l 691. Her assessment
can be verified using evidence on the
27
record.
model, there will be approximately
26
billion machineable,
Therefore, the capacity problem, which she identified as the primary
According
to witness Yacobucci’s
non-carrier
route flats in the Test Year.’
at
13
Excluding witness
“This number was calculated by summing machineable flat volumes for First-Class Mail,
Standard (A) Regular, Standard (A) Nonprofit, Periodicals Regular, and Periodicals Nonprofit
from USPS-LR-I-90. To do this, I used the method recommended by witness Yacobucci during
12
1
Kingsley’s thirty percent of flats that destinate in small zones leaves
2
approximately
3
in zones where the Postal Service plans to perform incoming secondary
4
on flat sorters.
5
nine billion machineable,
non-carrier
route flats that will destinate
sorting
The Postal Service will have more than enough capacity to sort all of
6
these flats to carrier route on flat sorters.
7
modification
6
utilization (USPS-T-l
9
approximately
In the Base Year, before the OCR
to FSM 881~~ and the Postal Service’s effort to increase FSM
0 at IO and 15) the Postal Service performed
five billion incoming secondary
sorts on FSM 881s.
10
107, Yrscrub.xls.
11
16 billion sorts in the Test Year.’
12
Kingsley’s confirmation
13
will be to automate
14
that at least half of the savings from Phase I will be from automating
15
secondary
16
AFSM 100 deployment
17
Kingsley, Tr. 5/1782; Tayman, Tr. 21314; Kingsley, Tr. W1660; DMA. et al.-T-l.
18
Phase I of the AFSM 100 deployment
USPS-LR-I-
will provide an additional
Based upon witness Tayman and witness
that the primary use of the Phase I AFSM 100 machines
incoming secondary
sorting and witness Kingsley’s estimate
sortation, witness But (DMA, et al.-T-l)
incoming
calculated that Phase I of the
will provide at least six billion incoming secondary
The 11 billion Test Year incoming secondary
sorts.
sorts (6 billion on AFSM
19
100s and 5 billion on FSM 881 s) that will be provided by FSM 881 s and AFSM
20
100s is more than enough to automate the incoming secondary
21
machineable
22
significant
23
the AFSM 100’s capability to combine zones, O’Tormey, Tr. 21/8370, the Postal
24
Service may even want to consider (as it did for the delivery point sequencing
sorting of all
flats that are not in small zones.g Given the capacity available, the
productivity gap between AFSM 100 sorting and manual sorting, and
of
cross-examination. Yacobucci, Tr. 5/1485-7. These calculations are made in MPA-LR-2,
worksheet “Total Volumes.”
‘Witness Kingsley noted that the FSM 661 modification significantly increased the volume of flats
that received their incoming secondary sort on a flat sorter. USPS-T-IO at 14.
‘1 calculated this figure using the method described by witness Kingsley. Kingsley, Tr. 511961. I
multiplied 166.5 Test Year machine-equivalents
(USPS-LR-I-63 at l-12) by 52 weeks/year, 6
days/week, 20 hours/day, and 15,000 pieces per hour. USPS-LR-I-90, worksheet
“Productivities,” footnote 7. Also, note that this completely ignores the Phase II AFSM
deployment.
‘Even if you use witness O’Tormey’s 13.5 billion Test Year AFSM 100 sorts, O’Tormey, Tr.
21/6371, which is approximately twenty percent less than the figure I developed, there will still be
enough capacity to sort all machineable flats that are not in small zones.
13
1
letters) sorting some flats that destinate in zones with less than ten routes in
2
automated
3
operations.
Kingsley, Tr. 5/l 980.
Second, on cross examination,
witness Kingsley’s explanation
of the
4
reason why the IS factor is fifty percent, rather than seventy percent, is
5
unsatisfactory.
Specifically,
she states:
6
I think that Mr. Yacobucci’s models [the USPS flats cost
model] take into account most of those situations, where it
is too small of a zone, we have reject rates, we have
nonmachinability issues, and some of it, again, is coverage
factors. We will not have AFSM 100s everywhere. There
are only 173 machines, and there are 250 processing
facilities. So we know that it is not going to be available to
cover every zone with 10 or more carrier routes. Kingsley,
Tr. 5/l 979.
7
6
9
10
11
12
13
14
15
16
17
While witness Kingsley is correct that the USPS flats cost model does take
16
these “situations”
into account, she is wrong that these situations should reduce
19
the IS factor to fifty percent.
Specifically,
20
represents
of flats by machine type that flow to a machine for
21
incoming secondary
22
USPS-T-25
23
without AFSM 100s and FSM 881 s do not “flow” to AFSM 100s and FSM 881 s
24
for incoming secondary
25
AFSM 100s and FSM 881 s but, based upon machine-specific
26
explicitly rejected from these machines.
“the percentages
as discussed
above, the IS factor
piece handlings that the machine actually processes.”
at 16. First, nonmachineable
flats and flats that destinate at facilities
sorting in the first place. Second, rejects do flow to
accept rates, are
27
28
29
4. FSM 1000 Productivities
The flats cost model uses actual productivities
but assumption-based
30
keying operations,
31
operations.
USPS-LR-I-90,
32
Specifically,
due to a lack of Base Year FSM 1000 BCR productivity data, the
33
Postal Service assumed that FSM 1000 BCR productivities
34
as FSM 881 BCR productivities.
35
productivity
worksheet
productivities
from MODS for FSM 1000
“Productivities,”
The inconsistent
and FSM 1000 BCR productivity
for FSM 1000 BCR
footnotes 4 and 6.
are exactly the same
treatment of FSM 1000 keying
results in a significant data
14
1
anomaly: the incoming secondary
FSM 1000 keying productivity,
2
hour, in the USPS flats cost model is higher than the incoming secondary
3
1000 BCR productivity,
4
“Productivities.”
798 pieces per hour. USPS-LR-I-90,
863 pieces per
FSM
worksheet
lo
5
Furthermore,
when compared
to all other FSM keying operations,
6
incoming secondary
7
high. As Table 5 shows, the incoming secondary
8
is approximately
9
productivities
FSM 1000 keying productivity
the
appears to be anomalously
FSM 1000 keying productivity
fifty percent higher than the average of all other keying
and thirty percent higher than the next-highest
10
treat FSM 1000 keying and BCR productivities
II
anomalies,
12
881 keying productivities.”
consistently
I assume that FSM 1000 keying productivities
productivity.
To
and to resolve these
are the same as FSM
13
Table 5. PFY 1998 Keying
14
Description
Productivities’
FSM 1000
FSM 881
Outgoing
594
664
ADC
543
531
Incoming Primary
599
556
Incoming Secondary
863
488
‘USPS-LR-I-90,
worksheet “Productivities.”
5. PostCom
Automation-Related
15
16
17
Cost Savings
To estimate the extent that the Postal Service’s incorrect assumptions
10
reduce automation-related
cost savings, I estimated automation-related
19
savings using the revised version of the flats cost model, MPA-LR-2,
20
developed jointly with witness Stralberg.
21
related cost savings based upon more reasonable
22
than those estimated by the USPS flats cost model.
cost
that I
As shown in Table 6, automationassumptions
are much higher
Further, based upon these
“This clearly is anomalous. According to MD FY 2000 MODS data, FSM 1000 BCR
productivities are about twice as high as FSM 1000 keying productivities.
Institutional Response
to PostComlUSPS-ST43-6,
Attachment 3.
15
1
improved automation-related
2
quality-related
3
percent passthrough
4
passthrough
cost savings estimates
savings), Moeller’s Basic Automation
and his 3/5-Digit Automation
(and still ignoring address
discount represents
a 128
discount is based upon a
of 204 percent.
5
6
Table 6. Comparison
of Automation-Related
Cost
Avoidances
7
Rate Category
Basic Nonautomation
Basic Automation
Basic Auto Savings
Moeller’s
Effective
Passthrough’
126%
3/5-Digit Nonautomation
3/5-Digit Automation
3/5 Auto Savings
8
9
10
Mail Processing
Cost Avoidance (Cents Per Piece)
PostCorn
USPS
Savings
Model
Model’
Difference
[I]
[2]
[3]=[1]-[2]
21.406
19.825
17.901
17.915
3.505
1.910
12.546
12.004
11.221
11.457
1.595
204%
1.325
0.547
0.778
‘Automation discounts, USPS-LR-I-166. worksheet “regval.” divided by PostCorn-modeled
automation cost avoidances.
‘USPS-T-25 at 5, Table H-4.
II
While witness Stralberg and I made several additional
12
improvements
13
the USPS flats cost model (for a detailed description
14
TW-T-l),
15
avoidances.
16
used in the revised model reduce automation-related
17
reduce the number of piece sorts that are required for presorted mail. As I
ia
describe in my testimony for MPA, these lower Test Year bundle breakage
19
assumptions,
20
reasonable.
to
of these improvements,
only one of these has a significant impact on automation-related
see
cost
Specifically, the lower Test Year bundle breakage assumptions
MPA-LR-2,
worksheet
cost savings because they
“Control Sheet,” are both appropriate
and
21
“Another
1000s.
option for resolving the anomaly would be using YTD FY 2000 productivites
for FSM
16
1
B. The Postal Service
2
Automation
3
Ignored
Requirements
As described
the Cost Savings
Pertaining
That Result Directly
to Address
From
Quality
by witness Lubenow, the high address quality of automation
4
flats results in significant
5
I show that address quality is higher for automation
6
mail; describe why better address quality results directly from automation
7
requirements;
8
explicitly models some of the cost savings that result from improved address
9
quality (e.g., higher accept rates), it ignores many others.
IO
(although hard to quantify) cost savings.
In this section,
mail than for nonautomation
and explain that while the Postal Service’s flats cost model
To set a context for this discussion,
(WA)
a recent Postal Service study of
11
undeliverable-as-addressed
mail, defined as “all mail that cannot be
12
delivered to the person or business at the address specified”, found that the total
13
cost for processing
14
Therefore,
UAA mail is about $1.5 billion. USPS-LR-I-82
at 30.
address quality does have a large impact on Postal Service costs.
15
16
17
1. High Address
Quality
Results
Directly
From Automation
Requirements
In his testimony, witness Lubenow explains why address quality is higher
18
for automation
flats than for nonautomation
19
two additional
points: (1) while the Postal Service’s Operations
20
Kingsley) has no data to prove that automation
21
than nonautomation
22
Kingsley agrees that higher address quality stems directly from automation
23
requirements.
24
flats. In this section, I simply make
witness (witness
flats have higher address quality
flats, she believes that this is the case; and (2) witness
First, when witness Kingsley was asked to confirm that there is a
25
difference in address quality between automation
flats and nonautomation
flats,
26
she states, “I would assume yes, but have no data to support.”
27
5118051807.
28
directly from automation
29
bear addresses that are sufficiently complete to allow matching to the current
30
USPS ZIP+4 File and must be matched using current CASS-certified
31
matching software to obtain the correct numeric ZIP+4 code. These are not
Kingsley, Tr.
Second, when asked whether higher address quality results
requirements,
she stated: “Automation
rate flats must
address
1
requirements
for non-automation
2
result in some differences
non-carrier
in address quality.”
route presort flats and this could
Id.
3
4
5
2. Address
Quality
Affects
Much More Than Reject Rates
In his testimony, witness Lubenow explains that the higher address quality
6
of automation
flats not only reduces the reject rate on flat sorters, but also
7
decreases
a
only explicitly modeled the cost savings from lower reject rates, stating that “LR-I-
9
90 accounts for any other mail processing
many other costs to the Postal Service.
costs caused by address problems via
10
the CRA cost adjustments.”
11
by address problems via the CRA cost adjustments
12
the purpose of determining
13
Yacobucci,
The Postal Service, however,
Tr. 511481. Accounting
automation-related
essentially
for costs caused
ignores them for
cost savings.
For the vast majority of Standard (A) Regular flats, most of the unit costs
14
added by the CRA adjustment,
15
WORKSHARING-RELATED
CRA COST” adjustment,
16
Standard
and therefore does not reflect cost differences
17
between rate categories.
18
ADJUSTMENT
19
cent, and this amount accounts for all worksharing-related
20
modeled, not just the cost of poor address quality.
21
“Cost Averaging”
22
poor address quality is unrelated to the reject rate, the Postal Service’s flats cost
23
model understates
(A) rate categories
3.484 cents, come from the “NOT
For 3/5Digit
presort flats, the “PROPORTIONAL
FACTOR” increases the automation
and “Scenario
Costs.”
automation-related
which is added to all
CRA
cost savings by less than 0.1
costs not explicitly
USPS-LR-I-90.
worksheets
Because a large portion of the cost of
cost savings pertaining to address quality.
24
25
C. Barcoded
26
Program
27
Flats Will Facilitate
the Postal Service’s
Flats Automation
As witness O’Tormey describes in his testimony, the Postal Service’s flats
28
automation/mechanization
efforts have been met with mixed results.
USPS-ST-
29
42 at 7. As shown in Table 7, these mediocre results can be traced directly to
18
the large decreases
in FSM productivity
shifted from manual processing
that have occurred as more flats have
to machine processing.”
Table 7. Flat Sorting
Productivity
(TPH in Thousands)’
Flat Sorting Machine
Fiscal Year
1993
746
1994
735
1995
719
1996
714
1997
670
1998
613
‘USPS-LR-I-263, BY96 variability dataset (unscrubbed).
6
7
a
Evidence in this case suggests that this decrease
least partially resulted from the challenges
presented
in productivity
by nonbarcoded
has at
flats:
9
10
11
12
13
14
15
16
17
ia
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
The introduction of the FMOCR presents an additional
challenge to the management of flats processing
operations. Ideally, all FSM 881 machineable flats should
be processed on the FMOCR. However, depending on the
mailbase being processed, the reality is that only 70% to
80% of the non-barcoded flats inducted will be read by the
FMOCR. which means that 20% tom30% of the flats
inducted will not be read by the FMOCR. Therefore,
occasionally there will not be enough processing time or
equipment available to key the non-reads coming from the
FMOCR on a multiposition flat sorling machine (MPFSM)
keying sot-l program and still meet service commitments.
At times, the high non-read rate of a particular mailbase
may make it less productive to process it through the
FMOCR and then key the rejects than it would be to key all
the flats the first time through. However, the much higher
throughput for FMOCR induction versus keyed induction
should eliminate the latter consideration in almost all
cases.
The ideal scenario for each facility is to maximize
automated flat processing and reduce keying operations to
a minimum. The bottom line, however, is that each facility
will need to evaluate FMOCR versus MPFSM processing
“Another data set in USPS-LR-I-263 shows that FY 99 FSM productivity was even lower than FY
96 FSM productivity. USPS-LR-I-263, LR263MPAxls. BY96 variability dataset (scrubbed).
19
for each processing operation, taking into consideration
site-specific productivities, machine availability, and
mailbase
readability.
USPS-LR-I-193.
Strategic
Improvement Guide for Flats Processing at 21.
6
While it is possible that the challenges
presented
by nonbarcoded
7
may not adversely impact the AFSM 100 deployment
a
have past FSM deployments,
9
even larger impact on the success of the AFSM 100 deployment:
flats
to the same extent as they
it is also possible that barcoded flats will have an
10
II
12
13
14
15
16
17
ia
19
20
21
22
23
24
25
26
27
28
29
[Dluring the initial deployment of the AFSM loo’s, the
potential volume of suitable mail will be greater than the
capacity of the machines to be deployed. We expect to
prioritize mail for processing on available AFSM 100’s to
achieve the best overall results. We have found that bar
coded mail generally meets our overall preparation
guidelines and processes more efficiently than non barcoded.
The AFSM has a throughput that is several times greater
than either the FSM 881 or the FSM 1000. By placing the
best mail available on the AFSM 100, we will maximize
throughput and minimize downtime, including downtime
that might result from jams that occur when inferior quality
mail is presented...
Increasing the volume of bar coded mail is a major part of
our strategy for flats, as it was for letters. Unger, Tr.
2118181-8182
30
31
Until we know the extent to which the presence of barcoded flats will affect
32
the AFSM 100 deployment,
I believe that it is prudent to continue encouraging
33
mailers to barcode flats with a discount at the level proposed by witness Moeller.
20
ATTACHMENT
1
A. SUMMARY
OF POSTCOM
PROPOSED
RATES
2
Regular Subclass
Automation
Letters
Basic
3digit
5-digit
Enhanced
Entered at destination:
BMC
SCF
DDU
Carrier Route
Entered at destination:
BMC
SCF
0.162
0.170
0.159
0.150
0.156
0.146
0.135
0.126
0.153
0.14,
0.130
0.121
0.146
0.134
0.123
0.114
0.156
0.137
0.131
0.153
0.132
0.126
0.146
0.125
0.119
0.062
0.041
0.035
0.041
0.035
0.062
0.041
0.035
0.062
0.041
0.035
0.564
0.564
0.564
0.470
0.470
0.470
0.444
0.444
0.444
0.411
0411
0.411
0.204
0.197
0.176
0.160
0.173
0.152
0.175
0.168
0.147
Flats (tx-rated)
0.271
Basic
315.digit
0.235
0.247
0.211
0.242
0.206
Flats (lb-rated)
per piece
0.135
Basic
0.099
315 digit
Non-letters
Basic
High-D
Saturation
(pc-mted)
0.162
0.161
0.155
0.135
0.099
0.135
0.099
Non-letters
per piece:
Basic
High-D
.sat”retio”
(,b.rated,
per pound:
Basic
3/5 digit
presort
Letter?3
Basic
3bdigit
0.661
0.661
0.547
0.547
0.521
0.521
Entered at destination:
BMC
SCF
0.246
0.229
0.222
0.205
DDU
per pound:
Basic
High-D
Saturation
0.217
0.200
Residual Shape Surcharge
Non-letters
Basic
Y6digit
(pc-rated)
0.315
0.262
Non-letters
per piece:
Basic
3/5 digit
(lb-rated)
per pound:
Basic
3/5 digit
0.291
0.236
0.266
0.233
0.179
0.126
0.179
0.126
0.179
0.126
0.661
0.661
0.547
0.547
0.521
0.521
Residual Shape Surcharge
Barcode Discount
DDU
L&ten
Basic
Auto
High-D
Baturation
0.160
0.030
Revised version of USPS-LR-I-166. wpl~comm.xls. Worksheet “Sum”
Revised version assumes 100 percent passthroughs of dropship cost savings
0.150